Explainable machine learning reveals key predictors of ICU mortality in COVID-19: functional outcomes and physiotherapy interventions in cardiovascular patients
Aprendizado de máquina explicável revela principais preditores de mortalidade em UTI por COVID-19: resultados funcionais e intervenções fisioterapêuticas em pacientes cardiovasculares
Jorge Amaral, Gabriel Ribeiro Cesario da Silva, Lucas Martins Thimoteo, Luis Felipe da Fonseca Reis, Chiara Andrade Silva, Arthur de Sá Ferreira
Abstract
Background: Cardiovascular diseases are a leading cause of mortality worldwide, with the COVID-19 pandemic intensifying their impact on intensive care unit outcomes. Functional impairments and reduced mobility among critically ill cardiovascular patients are linked to adverse outcomes, but their predictive value for mortality during intensive care hospitalization with COVID-19 remains underexplored. Aim: This study employs machine learning and explainable artificial intelligence to identify key predictors and optimize intervention strategies. Methods: This retrospective study analyzed data from 100 critically ill patients with cardiovascular diseases and COVID-19 admitted to a private hospital in Brazil. Functional assessments included scores of global muscle strength and mobility at admission. Machine learning models—Logistic Regression, Decision Tree, Random Forest, CatBoost, and Explainable Boosting Machine—were developed in Python. Interpretability analyses were performed using Shapley Additive Explanations to determine the most relevant predictors. Results: The best-performing model, Random Forest, achieved a sensitivity of 90.5% and specificity of 83.9%, with an accuracy of 0.92 (95% confidence interval: 0.83–1.00). Passive kinesiotherapy, restricted mobility, and invasive mechanical ventilation were strongly associated with in-hospital mortality, while active mobilizations such as walking and standing predicted better survival outcomes. Feature relevance analysis revealed critical feature interactions involving oxygenation levels, sedation, and mobility variables on mortality risks. Conclusion: Machine learning approaches identified predictors of mortality and reinforced the protective effects of active physiotherapy interventions for critically ill cardiovascular patients with COVID-19. These findings support the application of data-driven strategies to optimize rehabilitation practices in intensive care units and suggest the need for validation in larger populations.
Keywords
Resumo
Introdução: As doenças cardiovasculares são uma das principais causas de mortalidade em todo o mundo, com a pandemia de COVID-19 intensificando seu impacto nos desfechos em unidades de terapia intensiva. Comprometimentos funcionais e mobilidade reduzida em pacientes críticos com doença cardiovascular estão associados a desfechos adversos, mas seu valor preditivo para mortalidade durante a internação na unidade de terapia intensiva devido à COVID-19 ainda não foi explorado. Objetivo: Este estudo utiliza aprendizado de máquina e inteligência artificial explicável (XAI) para identificar os principais preditores e otimizar estratégias de intervenção. Métodos: Este estudo retrospectivo analisou dados de 100 pacientes críticos com doenças cardiovasculares e COVID-19 internados em um hospital privado no Brasil. As avaliações funcionais incluíram escores de força muscular global e mobilidade no momento da admissão. Modelos de aprendizado de máquina— Regressão Logística, Árvore de Decisão, Random Forest, CatBoost e Explainable Boosting Machine—foram desenvolvidos em Python. A interpretação dos modelos foi realizada com base na técnica de Shapley Additive Explanations para identificar os preditores mais relevantes. Resultados: O modelo com melhor desempenho, Random Forest, obteve uma sensibilidade de 90,5% e especificidade de 83,9%, com acurácia de 92% (intervalo de confiança de 95%: 83%–100%). A cinesioterapia passiva, mobilidade restrita e ventilação mecânica invasiva foram fortemente associadas à mortalidade hospitalar, enquanto mobilizações ativas, como caminhar e ficar em pé, previram melhores desfechos de sobrevivência. A análise de relevância das variáveis revelou interações críticas envolvendo níveis de oxigenação, sedação e métricas de mobilidade. Conclusão: Técnicas de aprendizado de identificaram preditores de mortalidade e reforçaram o efeito protetor das intervenções fisioterapêuticas ativas em pacientes críticos com doença cardiovascular e COVID-19. Esses achados apoiam adoção de estratégias de reabilitação orientadas por dados clínicos em unidades de terapia intensiva, com necessidade de validação em populações maiores.
Palavras-chave
Referências
1. Roth GA, Johnson C, Abajobir A, Abd-Allah F, Abera SF, Abyu G, et al. Global, regional, and national burden of cardiovascular diseases for 10 causes, 1990 to 2015. J Am Coll Cardiol. 2017;70(1):1-25. https://doi.org/10.1016/j.jacc.2017.04.052. PMid:28527533.
2. Brant LCC, Nascimento BR, Passos VMA, Duncan BB, Bensenõr IJM, Malta DC, et al. Variations and particularities in cardiovascular disease mortality in Brazil and Brazilian states in 1990 and 2015: estimates from the Global Burden of Disease. Rev Bras Epidemiol. 2017;20(20, suppl 1):116- 28. https://doi.org/10.1590/1980-5497201700050010. PMid:28658377.
3. Marinho F, de Azeredo Passos VM, Carvalho Malta D, Barboza França E, Abreu DMX, Araújo VEM, et al. Burden of disease in Brazil, 1990–2016: a systematic subnational analysis for the Global Burden of Disease Study 2016. Lancet. 2018;392(10149):760-75. https://doi.org/10.1016/ S0140-6736(18)31221-2. PMid:30037735.
4. Lotufo PA. The pace of reduction of cardiovascular mortality in Brazil (1990 to 2017) is slowing down. Sao Paulo Med J. 2019;137(1):3-5. https://doi.org/10.1590/1516- 3180.2018.1371090219. PMid:31116267.
5. Buddeke J, Bots ML, van Dis I, Liem A, Visseren FLJ, Vaartjes I. Trends in comorbidity in patients hospitalised for cardiovascular disease. Int J Cardiol. 2017;248:382-8. https:// doi.org/10.1016/j.ijcard.2017.06.106. PMid:28712563.
6. Aggarwal G, Cheruiyot I, Aggarwal S, Wong J, Lippi G, Lavie CJ, et al. Association of Cardiovascular Disease With Coronavirus Disease 2019 (COVID-19) severity: a meta-analysis. Curr Probl Cardiol. 2020;45(8):100617. https:// doi.org/10.1016/j.cpcardiol.2020.100617. PMid:32402515.
7. Bansal M. Cardiovascular disease and COVID-19. Diabetes Metab Syndr. 2020;14(3):247-50. https:// doi.org/10.1016/j.dsx.2020.03.013. PMid:32247212.
8. Clerkin KJ, Fried JA, Raikhelkar J, Sayer G, Griffin JM, Masoumi A, et al. COVID-19 and Cardiovascular Disease. Circulation. 2020;141(20):1648-55. https://doi.org/10.1161/ CIRCULATIONAHA.120.046941. PMid:32200663.
9. Creditor MC. Hazards of hospitalization of the elderly. Ann Intern Med. 1993;118(3):219-23. https://doi.org/10.7326/0003- 4819-118-3-199302010-00011. PMid:8417639.
10. Ferreira NA, Lopes AJ, Ferreira AS, Ntoumenopoulos G, Dias J, Guimaraes FS. Determination of functional prognosis in hospitalized patients following an intensive care admission. World J Crit Care Med. 2016;5(4):219-27. https:// doi.org/10.5492/wjccm.v5.i4.219. PMid:27896146.
11. Johnson JK, Lapin B, Green K, Stilphen M. Frequency of physical therapist intervention is associated with mobility status and disposition at hospital discharge for patients with COVID-19. Phys Ther. 2021;101(1):2189. https:// doi.org/10.1093/ptj/pzaa181. PMid:32986836.
12. Silva CA, Lopes AJ, Papathanasiou J, Reis LFF, Ferreira AS. Association of functional characteristics and physiotherapy with COVID-19 mortality in intensive care unit in inpatients with cardiovascular diseases. Medicina (Kaunas). 2022;58(6):823. https://doi.org/10.3390/medicina58060823. PMid:35744086.
13. Assaf D, Gutman Y, Neuman Y, Segal G, Amit S, Gefen-Halevi S, et al. Utilization of machine-learning models to accurately predict the risk for critical COVID-19. Intern Emerg Med. 2020;15(8):1435-43. https://doi.org/10.1007/s11739-020- 02475-0. PMid:32812204.
14. Akram T, Attique M, Gul S, Shahzad A, Altaf M, Naqvi SSR, et al. A novel framework for rapid diagnosis of COVID-19 on computed tomography scans. Pattern Anal Appl. 2021;24(3):951-64. https://doi.org/10.1007/s10044- 020-00950-0. PMid:33500681.
15. Scarpone C, Brinkmann ST, Große T, Sonnenwald D, Fuchs M, Walker BB. A multimethod approach for county-scale geospatial analysis of emerging infectious diseases: a cross-sectional case study of COVID-19 incidence in Germany. Int J Health Geogr. 2020;19(1):32. https://doi.org/10.1186/ s12942-020-00225-1. PMid:32791994.
16. Cheng FY, Joshi H, Tandon P, Freeman R, Reich DL, Mazumdar M, et al. Using Machine Learning to Predict ICU Transfer in Hospitalized COVID-19 Patients. J Clin Med. 2020;9(6):1668. https://doi.org/10.3390/jcm9061668. PMid:32492874.
17. Benchimol EI, Smeeth L, Guttmann A, Harron K, Moher D, Petersen I, et al. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) Statement. PLoS Med. 2015;12(10):e1001885. https:// doi.org/10.1371/journal.pmed.1001885. PMid:26440803.
18. Vanpee G, Hermans G, Segers J, Gosselink R. Assessment of limb muscle strength in critically Ill patients. Crit Care Med. 2014;42(3):701-11. https://doi.org/10.1097/ CCM.0000000000000030. PMid:24201180.
19. Kawaguchi YMF, Nawa RK, Figueiredo TB, Martins L, Pires- Neto RC, Maria Y, et al. Perme Intensive Care Unit Mobility Score and ICU Mobility Scale: translation into Portuguese and cross-cultural adaptation for use in Brazil. J Bras Pneumol. 2016;42(6):429-34. https://doi.org/10.1590/s1806- 37562015000000301. PMid:28117473.
20. Hastie T, Tibshirani R, Friedman J. The elements of statistical learning. New York: Springer New York; 2009. https:// doi.org/10.1007/978-0-387-84858-7.
21. Prokhorenkova L, Gusev G, Vorobev A, Dorogush AV, Gulin A. CatBoost: unbiased boosting with categorical features. In: Bengio S, Wallach H, Larochelle H, Grauman K, Cesa-Bianchi N, Garnett R, editors. Advances in Neural Information Processing Systems [Internet]. Curran Associates, Inc.; 2018 [citado em 2025 Jan 16]. Disponível em: https:// proceedings.neurips.cc/paper_files/paper/2018/file/14491 b756b3a51daac41c24863285549-Paper.pdf
22. Nori H, Jenkins S, Koch P, Caruana R. InterpretML: a unified framework for machine learning interpretability. arXiv. 2019:1909.09223. https://doi.org/10.48550/ arXiv.1909.09223.
23. Akiba T, Sano S, Yanase T, Ohta T, Koyama M. Optuna: a next-generation hyperparameter optimization framework. KDD 2019. 2019;2623-31. https://doi.org/10.1145/3292500.3330701.
24. Elreedy D, Atiya AF. A Comprehensive Analysis of Synthetic Minority Oversampling Technique (SMOTE) for handling class imbalance. Inf Sci. 2019;505:32-64. https:// doi.org/10.1016/j.ins.2019.07.070.
25. Miller T. Explanation in artificial intelligence: insights from the social sciences. Artif Intell. 2019;267:138. https:// doi.org/10.1016/j.artint.2018.07.007.
26. Kim B, Khanna R, Koyejo OO. Examples are not enough, learn to criticize! Criticism for Interpretability. In: Lee D, Sugiyama M, Luxburg U, Guyon I, Garnett R, editors. Advances in neural information processing systems [Internet]. New York: Curran Associates, Inc.; 2016 [citado em 2025 Jan 16]. Disponível em: https://proceedings.neurips.cc/paper_files/paper/2016/ file/5680522b8e2bb01943234bce7bf84534-Paper.pdf
27. Jutzeler CR, Bourguignon L, Weis CV, Tong B, Wong C, Rieck B, et al. Comorbidities, clinical signs and symptoms, laboratory findings, imaging features, treatment strategies, and outcomes in adult and pediatric patients with COVID-19: A systematic review and meta-analysis. Travel Med Infect Dis. 2020;37:101825. https:// doi.org/10.1016/j.tmaid.2020.101825. PMid:32763496.
28. Mehra MR, Desai SS, Kuy S, Henry TD, Patel AN. Cardiovascular disease, drug therapy, and mortality in Covid-19. N Engl J Med. 2020;382(25):e102. https://doi.org/10.1056/ NEJMoa2007621. PMid:32356626.
29. Liu J, Liu Z, Jiang W, Wang J, Zhu M, Song J, et al. Clinical predictors of COVID-19 disease progression and death: analysis of 214 hospitalised patients from Wuhan, China. Clin Respir J. 2021;15(3):293-309. https://doi.org/10.1111/ crj.13296. PMid:33090710.
30. Teixeira JJM, Bastos GCFC, de Souza ACL. Profile of hospitalization of the elderly. Rev Soc Bras Clin Med. 2017;15(1):15-20.
31. Rees EM, Nightingale ES, Jafari Y, Waterlow NR, Clifford SB, Pearson CA, et al. COVID-19 length of hospital stay: a systematic review and data synthesis. BMC Med. 2020;18(1):270. https://doi.org/10.1186/s12916-020-01726-3. PMid:32878619.
32. Bastos GAN, de Azambuja AZ, Polanczyk CA, Gräf DD, Zorzo IW, Maccari JG, et al. Características clínicas e preditores de ventilação mecânica em pacientes com COVID-19 hospitalizados no sul do país. Rev Bras Ter Intensiva. 2020;32(4):487-92. PMid:33263703.
33. Nishiga M, Wang DW, Han Y, Lewis DB, Wu JC. COVID-19 and cardiovascular disease: from basic mechanisms to clinical perspectives. Nat Rev Cardiol. 2020;17(9):543-58. https:// doi.org/10.1038/s41569-020-0413-9. PMid:32690910.
34. Attaway AH, Scheraga RG, Bhimraj A, Biehl M, Hatipoğ Lu U. Severe covid-19 pneumonia: pathogenesis and clinical management. BMJ. 2021;372:372. https://doi.org/10.1136/ bmj.n436. PMid:33692022.
35. Nasa P, Azoulay E, Khanna AK, Jain R, Gupta S, Javeri Y, et al. Expert consensus statements for the management of COVID-19-related acute respiratory failure using a Delphi method. Crit Care. 2021;25(1):106. https://doi.org/10.1186/ s13054-021-03491-y. PMid:33726819.
36. Bonorino KC, Cani KC. Mobilização precoce em tempos de COVID-19. Rev Bras Ter Intensiva. 2020;32(4):484-6. https:// doi.org/10.5935/0103-507X.20200086. PMid:33470350.
37. Mendez-Tellez PA, Needham DM. Early physical rehabilitation in the ICU and ventilator liberation. Respir Care. 2012;57(10):1663-9. https://doi.org/10.4187/ respcare.01931. PMid:23013903.
38. Kurtz P, Bastos LSL, Dantas LF, Zampieri FG, Soares M, Hamacher S, et al. Evolving changes in mortality of 13,301 critically ill adult patients with COVID-19 over 8 months. Intensive Care Med. 2021;47(5):538-48. https:// doi.org/10.1007/s00134-021-06388-0. PMid:33852032.
Submetido em:
16/01/2025
Aceito em:
09/12/2025
Facebook
LinkedIn
Mendeley
Reddit
Email